In Non-Orthogonal Multiple Access (NOMA)-enabled D2D cellular networks, geographically distributed devices compete for limited channels, making D2D pairing and channel selection a challenging but unexplored research problem. To this end, we formulate an optimization problem from the perspective of game theory with the objective of maximizing the data transmission rate while satisfying geographical and performance constraints, and theoretically prove that it is an Exact Potential Game (EPG). To find such solution, we propose an algorithm called Game-based D2D Pairing and Channel Selection (GDPCS) and prove its convergence. Also, the simulation experiments validate the effectiveness of GDPCS Algorithm and demonstrate its superior to existing state-of-the-art approaches.